AirDat system for ensuring TAMDAR data quality

AirDat is equipping commercial aircraft with a network of TAMDAR (Tropospheric Airborne Meteorological Data Reporting) sensors. The network uses communication satellites to relay atmospheric observations in near real time to a data processing center. It is important to maximize the amount of high quality atmospheric data, and minimize the amount of bad data, before it is delivered for use by operational meteorologists, aviation support systems, and forecasting models. Questionable data, or a decrease in the reliability of a sensor, must be quickly identified and acted upon. AirDat developed a quality assessment system to help achieve these goals. The system structure permits rapid automated responses to suspicious data and facilitates the flexible insight and problem identification skills that people provide.

Characteristics of the AirDat system used to monitor the quality of high volumes of meteorological data in a timely manner are described in this paper. It explores the type of problems detected by automated systems, the role of meteorologists and engineers in the quality assessment process, and how the system will evolve to incorporate faster automated responses to suspicious data.

Three key elements in the quality assurance process are considered:

1. “Real-Time Quality Filters” perform immediate assessment of incoming observations and determine what measurements are appropriate for distribution.

2. A separate system – named “Delta Hound” – performs an exhaustive automated analysis of observations. It looks for suspicious patterns, compares TAMDAR measurements to information available from reference sources, and provides tools to view the analysis results and baseline data.

3. “AirDat Personnel” use Delta Hound to evaluate the health of TAMDAR sensors and investigate data quality issues. They make high-level decisions about sensor reliability, data filtering, and repairs based on the Delta Hound analysis. Active human participation in the quality assessment process also provides the insight required for new problem detection algorithms and future enhancement of automated systems. Algorithms are initially implemented in Delta Hound; and, upon proving their worth, may form the basis of new real-time quality filters.